ABSTRACT In this application, we aim to directly address key gaps in our knowledge of NHL etiology and promote ?best practices? for future research. NHL etiologic research has largely been driven by pooled case-control studies, which have suggested a number of modifiable risk factors, but which also (i) may be influenced by unmeasured biases (survival, recall); (ii) cannot inform calculation of population attributable risk (ARp%); and (iii) have been limited by the lack of available tumor tissue for delineating molecular subtypes. Our study objective is to evaluate purported NHL risk factors using a pooled analysis of over 500,000 individuals followed for ~20 years from five cohort studies with >5000 incident NHLs. These cohorts have: (i) detailed longitudinal information on common modifiable risk factors queried in a similar manner over a similar period of time, and (ii) active tumor tissue collection efforts. In Aim 1, we will conduct a pooled time-dependent analysis in the prospective cohort studies to evaluate NHL risk factors consistently reported in case-control studies, assess the importance of exposure timing, and calculate the ARp% of confirmed risk factors. Specifically, we will evaluate five key purported modifiable risk factors: (i) young and usual adult body mass index (BMI); (ii) alcohol; (iii) sun exposure, (iv) smoking, and (v) menopausal hormone therapy; and two emerging risk factors of relevance to the immune response: (i) diabetes and (ii) chronic infections. In Aim 2, we will characterize and evaluate the importance of tumor molecular characteristics to these etiologic associations and ARp%. Specifically, unique gene expression patterns linked to survival in diffuse large B-cell lymphoma (DLBCL) and distinct gene mutations that denote disease progression in follicular lymphoma (FL) have been identified. ?Double hit? lymphomas have also been shown to have poor prognosis among multiple lymphoma subtypes. New multi-analyte gene expression assays developed for formalin-fixed paraffin-embedded tissues now make the delineation of these unique molecular signatures in epidemiologic studies possible. Because tissues archived from epidemiologic studies reflect only a subset of all cases, we will compare the epidemiologic and clinical characteristics of tissues retrieved from cohort studies to an unbiased series of tissues from Kaiser Permanente Southern California. This comparison will permit us to quantify survival bias in case-control study findings and potential bias in tissue ascertainment in the cohort-related tissues. The quantification of these biases in case ascertainment and tissue retrieval will enable us to interpret resulting Aim 1 risk associations and adjust ARp% calculations accordingly.